<jats:p> Abstract. Based on the principles of social comparison theory, the relative percentile (RP) method is an alternative approach to the measurement of psychological characteristics. It involves asking raters to explicitly estimate the percentage of a comparison group that they believe is lower than the target on a characteristic. This study explored the RP method for the measurement of personality. Specifically, we investigated the convergence of the RP with traditional (i.e., Likert-type) personality measures and the convergence between self- and observer reports. Both members of 142 Australian well-acquainted dyads rated themselves and their counterpart using the traditional Likert-type HEXACO-100 and a 25-item RP assessment of the HEXACO facets. Two weeks later, 78 participants completed the RP assessment again, allowing the assessment of test-retest reliability. The RP ratings showed mostly moderate reliability, though generally lower reliability than their corresponding traditional scales, and a relatively clear HEXACO factor structure. Furthermore, the RP ratings correlated significantly with the Likert-type ratings from the same rater (e.g., self–self) and with RP ratings from a different rater (i.e., self–observer), although convergence did vary by HEXACO domain. One potential issue with RP ratings, however, is that they mostly yielded Gaussian distributions, instead of the theoretically expected uniform distribution, which may suggest that it is challenging for respondents to estimate percentiles.